Jerry Hausman is a straight shooter. In this well written Journal of Economics Perspective piece, he presents a coherent critique of contingent valuation as a credible tool for valuing non-market goods. Long live hedonics?
Nothing wrong with hedonics. Nothing at all. Move along folks. And, weirdly, as it turns out, a number of studies find convergent validity between contingent valuation and hedonics (and other revealed preference methods). It must be pure coincidence. Luck. Happenstance.
Listen. Listen to me! Most economists who engage in stated preference method trashing engage in the weakest and emptiest sort of academic criticism. My favorite discussant comment received at a conference is "I wish there was a better way to do this" preceded by the classic "I don't like it when economists ask hypothetical questions." Classic. This seems to be enough of a criticism for many economists. Weak.
Three quick comments on Hausman's well written piece. He dismisses contingent valuation because of hypothetical bias, the WTP-WTA divergence and CVM studies that don't pass the Diamond-Hausman adding up test:
- When hypothetical bias exists, and I believe it does in many studies, it is systematic and readily adjusted for accuracy. In benefit cost analysis sensitivity analysis is a useful way of adjusting numbers that you know to be wrong (and there are plenty of these wrong numbers in most benefit cost analyses). With CVM estimates, there are a number of ways to adjust these upward biased WTP estimates downward (see Loomis) and a number of studies find that the adjustment procedures are accurate.
- Willingness to pay is less than willingness to accept because there are not many good market substitutes for many public goods and willingness to accept is not income constrained. If I'm selling my car I start the bidding high. Hausman doesn't reference Hanemann. Google scholar says that 970 others have.
- As far as I can tell, the only thing in economic theory to guide a scope test is that willingness to pay should be nondecreasing in quantity or quality (more is better and free disposal). More restrictive tests can be developed and the Diamond-Hausman test is a neat one, for sure, but failure of these tests should also cause one to wonder if the test is too restrictive. Hausman notes that Heberlein et al. admit that some contingent valuation studies fail to find scope. He doesn't mention that they also argue that scope shouldn't be the most critical validity test. I don't really agree with Heberlein et al. about that but Hausman leaves the impression that Heberlein et al. are happy to walk away from contingent valuation. They aren't.
I've conducted n = I don't know how many contingent valuation and other stated preference studies. Much of my research time these days is spent trying to figure out ways to deal with hypothetical bias in stated preference data. I imagine I know the problems of the method better than most. There are problems with contingent valuation, just like there are problems with revealed preference methods. But these problems are not intractable.
It is an insult to the research process and spirit of inquiry for an AEA journal to allow a top economist to "selectively review the contingent valuation literature" (p. 43), somehow miss well-cited studies that provide contrary evidence, and declare a method for assigning values "hopeless" (I'm truly surprised that the Journal of Economic Perspectives editors allowed that title). Proof by potentially outlying counter example isn't good social science.
Oh, and you guys should read this book.
Comments open. Straight shooters with coherent comments welcome.
*I was trying to take my time with the JEP symposium, focusing on the election by reading Deadspin's coverage and watching Tosh reruns, but Matt's post set me off. Sorry Matt, nothing personal: wrong place, wrong time. As the athletes say these days, it is what it is.